The Theory of Related-ivity: Segment X

The Theory of Related-ivity:

A History and Analysis of the Best Related Work Hugo Category

by Heather Rose Jones

(This is a serialized article exploring the history of the Best Related Work Hugo category in its various names and versions. If you’ve come in at the middle, start here.)

Contents

Part 3: Historic Trends

3.3 Category

3.3.1 Introduction

Part 3: Historic Trends

3.3 Category

3.3.1 Introduction

The method for assigning content Categories to works is detailed in the Category chapter in the Categorization Process section. This includes a discussion of how Supercategories are assigned. To recap: each work may be assigned multiple Category tags, based on the nature of the content, but only a single Supercategory tag is assigned, based on the preponderance of the subject matter.

This introductory chapter will compare high-level distribution of Categories with respect to various eras (all data, Non-Fiction, Related Book, Related Work) and subsets (Long List,[1] Finalists, Winners). The comparisons will be performed both on the Supercategory level and the Category level. After that, chapters will discuss each Category individually, grouped by overall popularity. More popular categories will analyze more factors, as they are more likely to show significant patterns, while less popular Categories will be discussed more anecdotally. Each Category will include a summary of its distribution data. Content-related data other than the content Categories will be discussed in the next section (Other Tags).

Category Distribution

The purpose of this initial analysis is to determine whether there have been shifts in the nature of the subject matter being nominated either due to changes in the category scope or simply over time. (These two factors can be difficult to distinguish unless a year-by-year comparison contains enough data for meaningful analysis.)

All Data

When examining the data set as a whole, there is a rough consistency of presence for 2 Supercategories (Analysis, Associated) across the process levels (all data, Finalists, Winners) while 2 become disproportionately more popular as the selection process is applied (People, Information), and 1 is disproportionately less popular (Image). The differences between levels are not large, however. Overall, the most popular Supercategory is Analysis (which includes works such as Reviews and Criticism), followed by Information (Reference works, histories), with People, Images, and Associated following after. One might think that the lower presence of Associated works could, to some degree, be attributed to the clustering of many of its component Categories in the Related Work era, but this is not the case (as seen below).

Table 6: Supercategory Distribution for All Eras Combined

When examined at the Category level, we can identify some clear distinctions in popularity. The “most popular” 3 categories each represent 14-15% of the complete data set (Art, Criticism, Essays). There is a gap to the top of the next most popular group at 9%. We can somewhat arbitrarily identify a “more popular” group at 6-9% of the total (Autobiography/Memoir/Letters, Biography, Craft, Fiction, History, Reference). The “less popular” group each represents 1-5% of the data (Conventions, Convention Publications, Graphic, Humor, Interviews, Journalism, Reviews, Science). This leaves a “least popular” group of categories each representing <1% of the data (Experience, Music, Photography, Poetry, Role Playing Game).

Many categories appear in roughly equivalent proportions at each level (all, Finalists, Winners) and in no case is there a difference of more than three percentage points between the popularity in the full data set and at the Finalist level. Comparing equivalence at the Winner level is tricker as a single Winner represents 2% of the whole. However, the following Categories stand out as either over- or under-represented. Disproportionate Winners: Autobiography/Memoir/Letters, History, Reference. Disproportionate non-Winners: Art, Criticism, Essays.

When considered in light of criticisms of some of the shifts in the category, the comprehensive data doesn’t support the assertion that works of marginal relevance are preventing more traditional works from being recognized. The Categories that win disproportionately are all what we might think of as “traditional” content, and even those “traditional” Categories that don’t win in proportion to their nomination (Art, Criticism, Essays) are among the most common winning Categories.

Table 7: Category Distribution for All Eras Combined

Non-Fiction

Has that distribution of nominations, Finalists, and Winners changed over time or with the changing definition of the Hugo category? The first step is to do a distribution analysis for each era of Best Related. (Keep in mind that, in the Non-Fiction era, Long List data is not available.) What’s striking in this era is the near equal distribution of Finalists across most Supercategories (with the exception of Associated), while Winners are not proportionate. Informational works make up nearly half of all Winners, with History and Reference works the main contributors. Works about People are only slightly over-represented, with Autobiography/Memoir/Letters providing the majority of the effect. Works focused on Images are startlingly underrepresented among Winners. (6% represents a single Winner.)

Table 8: Supercategory Distribution for the Non-Fiction Era

When individual categories are examined, the notable mismatch between Finalists and Winners is Essays (17% of the Finalists, no Winners).

Table 9: Category Distribution for the Non-Fiction Era

Related Book

In the Related Book era and following we have three levels of data to compare and find that representation of Supercategories in the Long List and Finalists is startlingly similar. Representation among Winners is also roughly similar with the following observations. Images are no longer at a disadvantage (indeed, are over-represented as Winners) and works about People continue to be popular while Information works have lost their edge.

Table 10: Supercategory Distribution for the Related Book Era

At a Category level, there’s nothing particularly surprising except to note that the rate of People Winners is anchored by Autobiography and its associated topics.[2]

Table 11: Category Distribution for the Related Book Era

Related Work

Proportions across the different levels continue to be roughly equivalent. The only really notable observation is the disproportionately low win rate for Analysis, though it’s still the most common type of Winner. (We noted this previously in the Non-Fiction era. Detailed cross-era comparisons are found later.)

Table 12: Supercategory Distribution for the Related Work Era[3]

Teasing out observations for the individual Categories, voters seem to prefer Autobiography over Biography and seem more willing to reward works of Fiction (an odd turn for an award that began by specifying Non-Fiction).

Table 13: Category Distribution for the Related Work Era

Long Lists

Moving to the question of whether nomination (and voting) patterns have shifted over time (whether due to the category changes or not) we will first compare Supercategory proportions in the Long List (which is not available for the Non-Fiction era). The largest change is for Images, which drop from representing a quarter of the Long List to only 4%. (There is an extensive discussion of this in the chapter on Art.) Otherwise, the changes seem less meaningful (and can simply represent adjustments for the absence of Images). But one lack of change is noteworthy. The remarkably consistent rate of Associated works before and after the change to the Related Work era suggests that there was no revolution in content at a general level as a result of the category expansion. Rather, the changes are at the Category level, such as a decrease in Graphic and the addition of Experience and Music. People might be nominating different types of Media formats (as previously discussed) but “marginal” subject matter is not taking up more space than it did when only Books were allowed.

Table 14: Comparison of Long List Supercategories Across Eras

Most of the changes in the proportion of specific Categories are minor (other than Art), but we might take note of increases in Biography, Essays, History, and Reviews (all quite “traditional” content) and decreases in Reference. Journalism is a new Category and clocks in at a solid 7%, unlike most new Categories which appear less frequently.

Table 15: Comparison of Long List Categories Across Eras

Finalists

When comparing Finalists, we have data for all three eras. The previously mentioned changes in the representation of Images are even more striking. An explanation would be welcome for both the increase in representation in the Related Book era then the near disappearance of the Supercategory in the Related Work era. Beyond that, People have been variable, but are as common now as they were at the start of the award, Analysis has been steadily increasing while Information decreases, and Associated content has grown to a smaller extent. Here we do see potential support for the impression that these less traditional content types are increasing in presence. These trends echo what we saw for the less complete Long List data.

Table 16: Comparison of Finalist Supercategories Across Eras

On an individual Category level, many of the shifts are minor or inconsistent across the three eras, but the following possible trends can be identified. Art (as already noted) crashes from being one more the most popular Categories to nearly disappearing. Autobiography seems to give way to Biography.[4] Criticism, History, and Reviews have gradually increased, while Reference works dropped significantly.[5]

Table 17: Comparison of Finalist Categories Across Eras

Winners

Here we see an echo of some of the shifts among Finalists. Works focused on Images disappear after having dominated the Related Book era. People as subjects fell in popularity in the Related Book era and remained steady after that. Analysis increased significantly in the Related Book era and rose only slightly more after that. Information fell drastically in the Related Book era and then remained steady. And Associated content disappeared in the Related Book era, then returned to the same rate as under Non-Fiction. Overall, it would be hard to conclude that there are any overall content trends that are associated specifically with the change to Related Work. Either the Related Book and Related Work numbers are similar or the Non-Fiction and Related Work numbers are similar. There is nothing that clearly distinguishes Related Work Winners from the types of content that won when the award was restricted to published Books. Even the lack of Image Winners under Related Work would not be so striking if not for the high rate of Winners in the middle era. Whatever effect the broadened scope of Related Work has had on Hugo voting, it hasn’t been with respect to the nature of the subject matter being elevated by voters.

Table 18: Comparison of Winner Supercategories Across Eras

There are few surprises when shifting to a Category-level analysis of Winners. Nine Categories have never supplied a Winner. Four Categories won in only one era (none of them under Related Book), but in 2 cases had more than one Winner in that era. Three Categories had Winners in 2 eras, with 2 of them having multiple Winners in a single era, but with all possible combinations of eras represented. The remaining 6 Categories (Autobiography, Biography, Craft, Criticism, History, and Reference) had Winners in all three eras, 2 of which had only a single Winner in each era.

The distribution of Winners doesn’t correlate exactly with the overall popularity among nominators but it comes close. Within the “most popular” Category group (based on overall presence in the data) only Criticism is a 3-era Winner, the others being 2-era Winners. The 6 Categories in the “more popular” group supplied 5 three-era Winners and one winning in only a single era. On the other side, the “less popular” group of 8 Categories had 4 winning members, some with multiple Winners, while the “least popular” group supplied no Winners. Other than large presence in the dataset being no guarantee of consistent wins (notably for Art and Essays), the correlation is solid.

Table 19: Comparison of Winner Categories Across Eras[6]

Distribution Data Within Categories

In each of the chapters discussing specific Categories, the following basic data is included, regardless of the number of works involved.

• The “percentage of years appearing” statistic is not discussed in detail in the individual chapters. It’s meant to give an indication of how consistently the Category has been present on the lists.
• The number of works with that Category tag and the percentage of the whole dataset they represent.

For each era:

• In how many years of the era (number and percentage) members of this Category appeared as Finalists.
• What percentage of the Finalists in that era they represent
• What percentage of Winners in that era have this Category tag

For the Related Book and Related Work eras:

• In how many years of the era (number and percentage) members of this Category appeared in the Long List
• What percentage of the Long List in that era they represent

For the larger categories, the percentage Long List and Finalist data is also shown in graphic form.

(Segment XI will cover Part 3 Historic Trends, Section 3.3 Category, Chapters under 3.3.2 Most Popular Categories.)